101

Earth-Observation Applications ofNavigation Satellites

P. SilvestrinEarth Observation Future Programmes Department, ESA Directorate of ApplicationProgrammes, ESTEC, Noordwijk, The Netherlands

IntroductionThe development of Earth-observationtechniques based on the use of navigationsignals can be traced back to the use of GPS in geodetic applications to provideastonishing precision, such as thedetermination to within a few centimetres ofthe relative positions of points on the Earth’ssurface separated by several thousandkilometres. This ultra-precise positioning offixed GPS receivers is now routinely used asa tool for studying the dynamics of theEarth’s crust (e.g. plate tectonics).

In the course of the development of therelated models and data-analysis techniques,new geophysical applications have emerged.Meticulous modelling work has shown how

phenomena that were initially considered asmeasurement perturbations for the originalapplication can later become the observationsof interest for new applications. For geodeticpositioning, the presence of water vapour(humidity) along the signal propagation path inthe troposphere (Fig. 1) is an unavoidablenuisance, because it causes a propagationdelay equivalent to a significant additional path(~ 0.4 m in humid regions). On the other hand,it is possible to derive this water-vapourcontent to about 1 kg/km

2 from the excess-

path estimates, which can be determined viathe geodetic analysis with about 1 mmaccuracy. This provides an opportunity toacquire data that are useful, for instance, fordetecting weather fronts in the regions wherethe receivers are deployed and, more generally,for improving weather forecasts if theinformation is properly processed in numericalweather-prediction systems. Several regionaland global networks of ground GPS receiversare in place today providing data for bothgeodetic and meteorological applications (see Table 1; not exhaustive). As an example, Figure 2 shows the German SAPOS (SAtelliten-POSitionierungdienst) network, data from which is used for various precise positioningapplications and will soon be used fortropospheric water-vapour determination also.

ESA contributes to the International GPSService (IGS) network with six receivers atvarious tracking stations, and with the routinedetermination of precise orbits and geodeticparameters at its European Space OperationsCentre (ESOC) in Darmstadt (cf. ‘SatelliteNavigation Using GPS’ in ESA Bulletin No. 90,May 1997, which also provides an overview ofGPS).

Additional information is provided by thecorrection for ionospheric effects realised withthe reception of navigation signals at twofrequencies (the ionosphere is a dispersivemedium and so the propagation paths dependon frequency). The information retrieved is thetotal content of electrons along the propagation

Satellite navigation systems, such as the US Global PositioningSystem (GPS) and its Russian counterpart GLONASS, were initiallyestablished to provide accurate positioning information. They have,however, also led to a great variety of unforeseen applications, many ofwhich are in fields that at first sight are unrelated to navigation. One ofthe most successful examples is the development of observationtechniques of interest for the Earth sciences.

On going developments, such as the advent of the Europeannavigation system Galileo, will enhance the use of these techniquesfor ocean and atmosphere remote sensing.

eo applications of navigation satellites

Table 1

Network No. of receivers Goals

EUREF, Europe > 100 Geodesy, tectonics,meteorology

GSI, Japan > 1000 Tectonics, meteorology

SAPOS, Germany > 150 Positioning, surveying, geodesy, meteorology

South California Net, USA > 250 Tectonics, meteorology

CORS + NOAA/FSL, USA > 100 Meteorology

International GPS Service > 200 Geodesy, tectonics, meteorology, ionosphere

Suominet > 100 Meteorology, ionosphere

102

path in the ionosphere. This is used in variousspace and ground applications, as well as instudies of ionospheric dynamics.

The spaceborne applicationsEarth-science applications are no longer limitedto the exploitation of ground receiver data. If ahigh-quality receiver is installed on a satellite inlow Earth orbit, it is possible to determine thatsatellite’s orbit to an accuracy of a fewcentimetres with respect to a network ofground receivers (fiducial stations). The flyingreceiver can measure very precisely andcontinuously the changes in the distances toseveral transmitters. From these data, it ispossible in a ground-processing step toreconstitute both the satellite trajectory andvarious geophysical parameters, such as thevariations in the Earth’s gravity field.

The performance of the technique has beendemonstrated on the US-French Topex-Poseidon mission dedicated to ocean altimetry,where an accuracy of 2–3 cm has beenachieved for the satellite height. The conceptcan be considered an extension of that ofgeodetic positioning of ground receivers: theorbit of a satellite is very smooth and highlypredictable (unless the satellite is manoeuvringor in a re-entry phase!) and this allows theorbiting receiver to provide measurements ofthe same quality as the ground receivers.Optimal estimation techniques are applied inthe ground-processing step and also provide

r bulletin 102 — may 2000 bull

Figure 1. Networks of receiverson the ground used to probethe troposphere’s humiditycontent and the ionosphere’selectron content

Figure 2. The SAPOS GPS receiver network of the German land surveying agencies

103

eo applications of navigation satellites

estimates of orbit perturbations of a geophysicalnature. The technique will therefore be used infuture missions aimed at improving ourknowledge of the Earth’s gravity field, such asthe Gravity and Steady-state Ocean CirculationExplorer (GOCE) selected in October 1999 asthe first ESA Core Earth Explorer mission forlaunch in 2005 (cf. ‘Probing the Earth fromSpace: the Aristoteles Mission’, ESA BulletinNo. 72, November 1992, and ‘The Gravity andSteady-State Ocean Circulation Explorer’, ESASP-1233 (1)). Several other altimetry and land-topography mapping missions also rely onGPS-based precise orbit determination.

Active atmospheric sounding by radiooccultationThe exploitation of the radio-occultationtechnique using navigation signals as thetransmitters of opportunity probably representsthe most important development for the Earthsciences. Well-known in Solar System missionswhere it has been applied since 1964 with thefirst observations of Mars’ atmosphere, thistechnique uses radio signals passing throughthe atmosphere in a so-called ‘limb-soundinggeometry’ to sense atmospheric properties.When applied to the Earth, it can provide dataof great value for meteorology and climatology,particularly because of its excellent geographicalcoverage, accuracy, vertical resolution and all-weather capability.

Our understanding of the Earth’s atmosphereand our capabilities for modelling and forecastingits changes are currently hampered by ourlimited knowledge of the temperature andhumidity fields. Reliable weather forecastsrequire an accurate description of the initialstate of the atmosphere, because small initial-state errors can grow rapidly in the forecast. Onlonger time scales, climate models predict aglobal surface temperature increase of about 1 K (1°C) over the next 20 to 30 years,accompanied by an increase in humidity and acooling in the upper part of the atmosphere,namely the stratosphere, extending from about15 to 50 km height. Climate studies dependcritically on the availability of accurate

observations of long-term consistency, which iswhat the radio-occultation measurements fromnavigation satellites can provide.

As navigation signals pass through theatmosphere, either through a descendingpropagation path (so-called ‘set event’) or anascending one (‘rise event’) when observed byan orbiting instrument, they are retarded andrefracted (bent) through an angle determinedby the gradients of refractivity along the path,as depicted in Figure 3. The main refractioneffect is caused by the vertical gradient ofrefractivity in the lowest atmospheric layercrossed by the path. The effect results in anexcess path and in a decrease in the observedsignal Doppler shift compared to a path invacuum. The path can be precisely determinedfrom the phase and amplitude measurementsof the received signals and used to determinethe refraction-angle profile through a rise or setevent. The refractivity gradients depend on the gradients of air density (and ultimatelytemperature), humidity and electron content. Inthe region above about 8 km (upper troposphereand stratosphere, up to 40 – 50 km), thehumidity is negligible and the refractivity is duemainly to vertical temperature gradients, so thattemperature profiles can be retrieved. Belowthis, the humidity effect dominates and, if atemperature estimate of moderate accuracy isavailable, the humidity profile can be retrieved.Vertical profiles of electron density in theionosphere can also be reconstituted.

Figure 3. Principle of radiooccultation using navigationsignals: the signal isreceived on a low-Earthorbiting satellite, and thepropagation path isobserved to bend becauseof atmospheric refraction

Figure 4. Typical excess-path and Doppler-shift

profiles through a radio-occultation event

The first proposal to use GPS signals for radiooccultation measurements was advanced byRussian scientists in 1987, shortly followed bya US proposal that led to a proof-of-conceptwith the GPS/MET experiment launched in April 1995 on the MicroLab-1 small satellite.Following extensive GPS/MET data analysisand research into the retrieval process, asignificant part of which was conducted inEurope through ESA studies, it is now widelyaccepted that radio occultation with navigationsignals can provide useful profiles of density,pressure and temperature in the middleatmosphere and the colder troposphere, aswell as humidity profiles in the tropical and mid-latitude regions of the troposphere. Thefeatures that make this technique so interestingfor meteorology and climate studies include theglobal coverage (about 1000 profiles per daycan be obtained for each instrument on a polarlow Earth orbiter, with 48 navigation satellites),the all-weather capability (no blockage byclouds, unlike classical satellite sounders), thehigh vertical resolution (varying from some 1.5 km in the stratosphere to some 0.2 km inthe troposphere), the retrieved temperatureaccuracy (about 1 K up to 35 km), and the long-term consistency. This latter quality stems fromthe fact that the measurements are essentiallytime-interval observations, which can bereferred to fundamental metrological standards(atomic time).

In addition to measuring the precise phase andamplitude of the carrier signal from eachnavigation satellite occulted by the atmosphere,the instrument also measures other signals thatdo not cross the atmosphere so as to determineprecisely its own velocity and position. Theseare needed to compute the (large) Doppler shiftcaused by orbital motion and derive the excess

Doppler caused by the propagation in theatmosphere. The required accuracy, particularlyfor the along-path velocity component, is high(order 0.1 mm/s), which makes it mandatory touse differential positioning techniques, supportedby a global ground network of receivers. Theuse of differential techniques also allows thecorrection of transmitter and receiver oscillatorerrors, which might otherwise be confused withthe atmospheric contribution. When thegradients of refractivity in the atmosphere aresmall over spatial scales of the order of thesignal wavelength (0.2 m), the processing canbe based on geometric optics, so that thepropagation paths can be treated as well-localised rays. The excess-Doppler profile canbe transformed into a refraction-angle profile,which in turn enables one to derive a refraction-index profile.

The atmospheric parameters of interest arelinked by a simple mathematical relation to therefraction index, which enables one toreconstitute them with the help of the knowngas and hydrostatic equilibrium laws. Therefraction-angle profiles can also be useddirectly in numerical weather-prediction modelsin order to combine them with other atmosphericmeasurements in an optimal estimationapproach, which is routinely done in operationalmeteorology. These prediction models canprovide temperature estimates with goodaccuracy (2 – 3 K) over the humid regions,where radio-occultation data can therefore beused to retrieve humidity with about a 10 – 20% accuracy.

Figure 4 shows a typical profile of the excesspath. This is clearly a substantial effect ifcompared to the millimetric accuracy withwhich one can measure it. The profile has beengenerated with a complex software simulatordeveloped under an ESA study in order toanalyse the end-to-end performance of thetechnique, and currently in use at variousEuropean institutes and companies. Figure 5shows an example of temperature and water-vapour pressure retrievals from actual datacollected in the GPS-MET experiment andcompared to similar retrievals from numericalweather-prediction systems using a large set ofconventional measurement systems (radiosondes,satellite radiometers, etc.). The total refractionangle reaches about 1 deg at grazing incidenceand can be determined from the data to anaccuracy of better than 0.01% of a degree.Because of the limb-sounding geometry, thehorizontal resolution of the measurements isbetween 100 and 300 km, which is sufficientconsidering the atmosphere’s stratifiedstructure and assuming proper data treatmentin weather and climate models.

r bulletin 102 — may 2000 bull

104

Figure 6. Illustration ofmultiple propagation effectsin radio occultation causedby large atmosphericrefractivity gradients. Theeffects are removed byprocessing the receiveddata so as to propagate thepaths backward up to a(virtual) auxiliary plane

Figure 5. Example oftemperature and water-vapour-pressure retrievalsfrom data collected in theGPS-MET experiment,showing the agreement with similar retrievals fromnumerical weather-prediction systems that usea large set of conventionalmeasurement systems. TheGPS_Tdry retrieval neglectsthe presence of humidityand results in anunderestimate oftemperature(Data courtesy of the UCARGPS-MET Project, theNational Center forEnvironmental Prediction(NCEP, USA) and theEuropean Centre for Medium-Range Weather Forecasts(ECMWF, UK))

method requires one to measure also theamplitude of the received signal. An addedadvantage of the method is that it allows thevertical resolution to be sharpened further andmakes it possible to observe vertical features inthe atmosphere with sizes below 100 m. All of the retrieval methods have been verified by means of the end-to-end performancesimulation software mentioned and will be usedin the data processing for forthcoming missions.

Other studies have addressed the accuracy ofthe GPS/MET retrievals compared with othersensors, e.g. radiosondes providing in-situmeasurements of comparable accuracy, andwith various weather-prediction models. Theresults show that temperature errors are belowor close to 1 K, although the comparisonbecomes difficult for regions where the errors ofexisting sensors and models are above thisvalue. This is the case for a large part of theSouthern Hemisphere, where the limitations inthe existing data are particularly strong.

Often the geometric (ray) optics description isno longer valid in humid regions, because thecomplex geometry of the humidity field cancause rapid fluctuations in refractivity. Suchfluctuations cause multiple propagation pathsand significant diffraction effects. In this case,the probed region behaves like a focussingrather than a defocussing lens, and the sensormeasures a rapidly varying signal. In terms ofrays, it is as if the instrument is measuring theresult of the interference of an unknownnumber of rays (Fig. 6). To handle such effectscorrectly, the data processing must be basedon wave optics (physical optics). A satisfactorymethod has been elaborated in the course ofvarious studies, based on the concept that therays can be ‘disentangled’ by propagating thereceived signal backwards until it isreconstructed on a virtual surface (auxiliaryplane in Fig. 6) where the effects are negligible.Once this is achieved, the retrieval can becarried out as before, starting with thederivation of the excess Doppler shift. The

eo applications of navigation satellites

105

Alt

itud

e, k

m

Temperature, ˚C Water Vapour Pressure, mb

Figure 7. Anechoic chambertesting of the GRAS antennabreadboard developedwithin the EOPP and GSTPprogrammes(courtesy of Saab EricssonSpace)

Instrument development Navigation sensors, even if space-qualified,cannot be used as instruments for theapplications described, since substantiallyhigher performances and different modes ofoperation are needed. For this reason, thedevelopment of suitable instruments wasinitiated in 1994 within ESA’s Earth ObservationPreparatory Programme. Relevant differenceswith respect to navigation equipment include: alarger number of receiving channels, since bothpositioning and sounding must be supported;low-noise measurements at high data rates; anultra-stable oscillator; high-gain beam-shapedantennas, to compensate for losses incurred inthe atmospheric propagation; improvedsensitivity to observe the weak signals during a

tropospheric occultation,including a so-called ‘open-loop’ mode where an on-board atmospheric modelaids the instrumentoperation when closed-loopphase locking of the signalis no longer possible. TheGNSS Receiver forAtmospheric Sounding(GRAS) has beendeveloped to provide radio-occultation measurementsin an operational way.Initially proposed in 1996 asthe instrument for aprospective constellation ofmicro-satellites (cf. ‘TheAtmospheric ProfilingMission’, ESA SP-1196 (7)),it is now being implementedon the Metop mission

following a request by Eumetsat to include it inits meteorological payload. It is worthmentioning that the experience gained withGRAS has also given European industry theopportunity to win important developmentcontracts for similar sensors in the US market.Recently, additional activities have been startedto develop a miniaturised version of GRAS forthe Atmospheric Climate Experiment (ACE)mission, selected in 1999 as a potential EarthExplorer Opportunity mission and currentlybeing studied at Phase-A level, as well as forother flight opportunities.

Emerging applicationsThe Earth-observation applications of navigationsystems could soon extend to ocean remotesensing, following various proposals elaboratedsince the late eighties regarding the exploitationof navigation signals reflected at the seasurface. The goals are mainly observation ofthe sea state (waves) and surface winds, andocean altimetry. While some successful proof-

of-concept airborne experiments have beenconducted (see ESA Bulletin No. 101, p. 133),studies and further experiments are nowunderway to assess the feasibility andusefulness of a spaceborne system. If these areconfirmed, benefits similar to those of theatmospheric-sounding application could beachieved via this new use of navigationsatellites as transmitters of opportunity. Theseinclude the reduced implementation effort,because of the receive-only instrumentation,and a substantial data return because of thelarge number of transmitters. Improved spatialand temporal sampling of both ocean andatmospheric conditions could then beachieved.

Future prospectsThe advent of the European navigation systemGalileo, and the planned GPS enhancements,represent promising developments for all of theapplications described here, mainly because ofthe anticipated increases in the number oftransmitting satellites, in signal power, and inthe number of carrier frequencies. Futureinstrument developments will certainly takeadvantage of these advances.

On the user side, while the operationalmeteorology community is already working onthe exploitation of ground-based water-vapourand radio-occultation data, climate scientistsare becoming increasingly interested in theprospective of a long-term climate observingsystem based on principles very different fromthose of the current instruments and able toprovide improved long-term data consistency.Proposals such the ACE mission mentionedabove and similar ones in the USA are driven bythis vision. Considering how quickly the initialconcepts for using navigation satellites forEarth observation have moved from a pureresearch scenario to one driven by operationalapplications, we can expect a very healthyfuture for this young domain of spacetechniques.

AcknowledgementThe current European leadership in theexploitation of the radio-occultation techniqueis the result of the work and dedication of many ESA contractors, including the DanishMeteorological Institute, the University of Graz(A), the Max-Planck Institute for Meteorology(D), the Aeronomy Service of CNRS (F), theCatalan Institute for Space Studies (E), the UKMeteorological Office, the University of Leeds(UK), Austrian Aerospace, Saab EricssonSpace (S) and Terma (DK). The support of theESA Earth Science Division is also gratefullyacknowledged. r

r bulletin 102 — may 2000 bull

106